RismadarVoice Reporters
September 4, 2026
Technology companies are facing criticism over reports that millions of physical books, including some rare and out-of-print titles, have been purchased, scanned and destroyed in the process of collecting data to train artificial intelligence models.
The controversy centres on a process known as “destructive scanning”, where books are cut apart to allow their pages to be scanned and converted into digital data for use in training large language models. Reports and court documents have revealed that some AI companies acquired large collections of books for this purpose, sparking concerns among authors, archivists and book preservation groups.
Supporters of the practice argue that physical books provide high-quality human-written material that can improve AI systems, particularly as companies seek reliable training data amid increasing concerns over copyright and the spread of AI-generated content online.
However, critics say destroying physical copies, especially rare or historically significant works, risks damaging cultural heritage and removing valuable resources that cannot easily be replaced. Some booksellers have also reported unusual bulk purchases of obscure titles, raising concerns about the impact on rare book collections.
The debate has also raised legal questions over whether companies can use purchased books to train AI systems without additional permission from authors and publishers. While some legal decisions have considered certain forms of book scanning for AI training under fair use principles, copyright disputes involving AI companies remain ongoing.
Technology firms have defended their efforts as part of developing improved AI services, while critics continue to call for greater transparency and stronger safeguards to protect literary works and historical materials.
Al Jazeera’s Linh Nguyen examined the controversy surrounding the reported destruction of books and the wider debate over whether the practice represents a significant threat to global literary preservation.
